import pickle
from util.load_mnist import load_mnist_train_XY
import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import norm
from sklearn.metrics import pairwise_distances
x, y = load_mnist_train_XY()
'''
x_r = load_robust_output()
x_nr = load_nonrobust_output()
'''
with open('../robust_morph.bin', 'rb') as infile:
x_r = pickle.load(infile)
with open('../nonrobust_morph.bin', 'rb') as infile:
x_nr = pickle.load(infile)
numSamples = x.shape[0]
x = x.reshape((numSamples, -1)).astype(np.float) / 255.0
x_r = x_r.reshape((numSamples, -1))
x_nr = x_nr.reshape((numSamples, -1))
locations = [np.where(y == i)[0] for i in range(10)]
result = []
for i in range(10):
sampleMat = x[locations[i], ...]
import warnings
from util.load_mnist import load_mnist_train_XY, preprocess_mnist_data
from sklearn.model_selection import train_test_split
warnings.filterwarnings(action='ignore', category=FutureWarning)
import pickle
import tensorflow as tf
_, y = preprocess_mnist_data(*load_mnist_train_XY())
#data = load_robust_output()
#data = load_nonrobust_output()
with open('../robust_morph.bin', 'rb') as infile:
data = pickle.load(infile)
train_X, test_X, train_Y, test_Y = train_test_split(data, y, test_size=0.1)
def get_new_network():
network = tf.keras.Sequential([
tf.keras.layers.Conv2D(32, (3, 3),
input_shape=train_X.shape[1:],
activation='relu'),
tf.keras.layers.Conv2D(64, (3, 3), activation='relu'),
tf.keras.layers.MaxPooling2D((2, 2)),
tf.keras.layers.Dropout(0.4),
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(100, activation='relu', name="FC"),
tf.keras.layers.Dropout(0.4),
tf.keras.layers.Dense(
10, activation='softmax') # no activation for the last layer
])
import subprocess
from util.load_mnist import load_mnist_train_XY
import time
import re
import threading
from queue import Queue
train_X, _ = load_mnist_train_XY()
concurrency = 5
readLineTimeout = 0.1
def enqueue_output(id, fileobj, queue):
for line in iter(fileobj.readline, b''):
queue.put((id, line))
def reconstruct_robust_dataset(left=0, right=-1):
if right == -1:
right = train_X.shape[0]
assert right > left
numSamples = right - left
taskStep = numSamples // concurrency
params = []
for i in range(concurrency):
params.append([